Ceramic diversity, random copying, and tests for selectivity in ceramic production

This paper examines the evidence for selective reproduction of ceramic types based on their evolving frequency distribution, and on correlations of the abundance of types with underlying functional characteristics. How can we tell if a variant has been preferentially selected or discriminated against, and become more or less frequent than would be expected under pure drift (i.e. random copying, with the compounding over time of proportionate sampling error)? We examined a database of Hittite ceramic bowl types from two successive Phases of occupation of the Upper City of Boğazköy–Hattusa, capital of the Hittite empire and the largest Bronze Age settlement in Turkey. We applied two tests used in genetics to assess departures from neutrality, to assess selectivity in rates of reproduction of ceramic bowl types (the Ewens-Watterson and Slatkin's Exact tests). We also examined the effects of ceramic fabric and vessel dimensions on changes in vessel abundance between the two Phases, using regression analysis. We found that while the frequency distribution of rim sherds did not in itself enable us to reject the null hypothesis of random copying, closer examination of the characteristics of these types enabled us to recognize latent dimensions of functional variability (including ware type and bowl diameter) that had demonstrably been the subject of selective decision-making by the potters. The present case study suggests that we should be wary of applying the neutral model from genetics uncritically in archaeology, because it is much harder to prove that the cultural traits whose frequencies are being modelled are genuinely functionally equivalent (as that model requires).